Genotoxicity testing is an important component of toxicity assessment. As illustrated by the European registration, evaluation, authorization, and restriction of chemicals (REACH) directive, it concerns all the chemicals used in industry. The commonly used in vivo mammalian tests appear to be ill adapted to tackle the large compound sets involved, due to throughput, cost, and ethical issues. The somatic mutation and recombination test (SMART) represents a more scalable alternative, since it uses Drosophila, which develops faster and requires less infrastructure. Despite these advantages, the manual scoring of the hairs on Drosophila wings required for the SMART limits its usage. To overcome this limitation, we have developed an automated SMART readout. It consists of automated imaging, followed by an image analysis pipeline that measures individual wing genotoxicity scores. Finally, we have developed a wing score-based dose-dependency approach that can provide genotoxicity profiles. We have validated our method using 6 compounds, obtaining profiles almost identical to those obtained from manual measures, even for low-genotoxicity compounds such as urethane. The automated SMART, with its faster and more reliable readout, fulfills the need for a high-throughput in vivo test. The flexible imaging strategy we describe and the analysis tools we provide should facilitate the optimization and dissemination of our methods.

We developed an automated mutant hair counting system in Drosophila wing images. In our previous work [2], we developed the image acquisition method using multi-focused image stack, hair separation method into upper and lower hair, and hair detection method using template matching. In this work, the hair detection and mutant classification algorithms are enhanced and extended to 3D, and the wing area segmentation is newly developed.

To test the genotoxicity of drug candidates in the process of drug discovery, The Drosophila wing somatic mutation and recombination test (SMART)[3][1] is used, which analyzes the development frequency of mutant hairs of Drosophila. Because the current mutant hair counting method is totally manual work, in which humans search and count the mutant hairs by their eyes through the microscope, its speed and accuracy is limited. Our goal is to culture the mutant Drosophila using possibly genotoxic chemicals, to take the image of Drosophila wing hair, and to develop an automated system that counts mutant hairs and tells the genotoxicity of the chemicals. In this work, we explain the image acquisition method using multi-focused image stack, hair separation method into upper and lower hair, and hair detection method using template matching.